AI Adds New Life in Old Photos

Nostalgia website MyHeritage has launched a new service that lets you create realistic animations of faces in photos.

The AI-based service called Deep Nostalgia, launched last week, is free to test and is very accurate in describing what someone would look like if they were captured on video. His eyes sparkle, his head moves, and his mouth form a smile.

To create deep nostalgia, MyHeritage teamed up with D-ID, an Israeli company focused on deep learning and synthetic media.

Creepy?

MyHeritage explained that Deep Nostalgia uses several pre-recorded directing videos prepared by the nostalgia company, which direct the motions in the animation and consist of a series of human gestures.

A preferred driver is automatically selected for each face based on orientation and then applied seamlessly to the photo. For best results, photos before animation are enhanced with MyHeritage Photo Enhancer, which blurs faces at low resolution and increases their resolution.

Users can animate multiple photos for free regardless of the number of faces in the photo. Continued use also requires a subscription. Video animation can be downloaded as an MP4 file and shared on social media.

While some people like to breathe new life into old photos, they may find using them a little daunting.

Tip of the iceberg

“There is no one-size-fits-all answer to whether this kind of deep counterfeiting is scary or funny,” adds Jean-Claude Goldenstein, CEO of CREOpoint, a San Francisco brand, and reputation protection provider.

There are many interesting uses for technology in Deep Nostalgia, such as video editing and video conferencing.

Deep False Threat

While there are many valuable uses for synthetic media, Castro said deep counterfeiting is a serious problem. “But this needs to be addressed through legislation that addresses specific harms, such as inaccurate electoral information, harassment or other harmful acts, such as spreading false nudity on someone without their consent,” he said.

Account Management Tool

In a Litan research paper, he predicted that by 2023, 20% of successful account control attacks will use deep spoofing to create social users to provide sensitive data or transfer money to criminal accounts.

Criminals take control of user accounts to gain access to confidential information and intellectual property and to steal money from financial accounts. Most money and information theft start with managing user accounts, as criminals use inadequate security controls that often protect those accounts.

Translate »